Another checksum mismatch

Hello

I just got this

   $ svn up ../svn
   U ../svn/subversion/tests/clients/cmdline/prop_tests.py
   ../svn/subversion/libsvn_ra_dav/fetch.c:687: (apr_err=155017)
   svn: Working copy text base is corrupt
   svn: could not save file
   ../svn/subversion/libsvn_wc/update_editor.c:1235: (apr_err=155017)
   svn: apply_textdelta: checksum mismatch for '../svn/subversion/tests/clients/cmdline/.svn/text-base/svnlook_tests.py.svn-base':
      recorded checksum: 7802170860e82408ffffffffa0cc0508
      actual checksum (hex): 325d0eeb6dc50c2dc123b8dc49638627
      actual checksum (base64): Ml0O623FDC3BI7jcSWOGJw==

I'm using r5371 and I have

   $ svn info svnlook_tests.py
   Path: svnlook_tests.py
   Name: svnlook_tests.py
   Url: http://svn.collab.net:81/repos/svn/trunk/subversion/tests/clients/cmdline/svnlook_tests.py
   Revision: 5377
   Node Kind: file
   Schedule: normal
   Last Changed Author: philip
   Last Changed Rev: 4400
   Last Changed Date: 2003-01-16 01:38:22 +0000 (Thu, 16 Jan 2003)
   Text Last Updated: 2003-01-14 23:41:19 +0000 (Tue, 14 Jan 2003)
   Properties Last Updated: 2003-01-15 14:11:50 +0000 (Wed, 15 Jan 2003)
   Checksum: 7802170860e82408ffffffffa0cc0508

If I do

   $ svn cat -r4400 svnlook_tests.py > zz
   $ diff .svn/text-base/svnlook_tests.py.svn-base zz

I see that my text-base is correct, so it really is the checksum in
the entries file that is wrong. The checksum in the entries file
doesn't appear to match that of any revision of the file. I'd go as
far as to say that those four consecutive ff bytes look like
unitialised memory.

From the log messages it appears that when the file last changed, in
r4400, it was a commit of mine that only set svn:executable. Neither
the working copy or text-base appears to have changed for some time

   $ ls -l .svn/text-base/svnlook_tests.py.svn-base svnlook_tests.py
-r--r--r-- 1 pm pm 4901 Jan 14 23:50 .svn/text-base/svnlook_tests.py.svn-base
-rwxr-xr-x 1 pm pm 4901 Jan 14 23:41 svnlook_tests.py

The Jan 14 date appears to correspond to r4390, another one of my
commits, which is the last time the text of the file changed.

I have so far failed to produce a test case that results in a corrupt
checksum, I'm tempted to build a circa 4400 client to see if I can
reproduce it with that.

-- 
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org